TDA Progress Report 42-96 October - December 1988

Radio Frequency Interference Survey Over the 1 .O-10.4 GHz Frequency Range at the Goldstone-Venus

Development Station S. Gulkis and E. T. Olsen

M. J. Klein and E. B. Jackson Space Physics and Astrophysics Section

Telecommunications and Data Acquisition Science Office

\ \

The results o f a low sensitivity Radio Frequency Interference (RFI) survey carried out at the Venus Station of the Goldstone Communications Complex are reported. The data cover the spectral range from 1 GHz to 10.4 GHz with a 10-kHz instantaneous band- width. Frequency and power levels were observed using a sweep-frequency spectrum analyzer connected to a 1-m diameter antenna pointed at zenith. The survey was con- ducted from February 16,1987 through February 24, 1987.

1. Introduction Plans are currently being made at the Jet Propulsion Labo-

ratory to carry out a comprehensive, ali-sky search for radio signals of extraterrestrial origin. The Search for Extraterrestrial Intelligence (SETI) survey will employ the Goldstone Commu- nication Complex near Barstow, California and other sites in the northern and southern hemispheres. The principal param- eters of this survey are given in Table 1. In preparation for this search, JPL has constructed a radio spectrum surveillance sys- tem (RSSS) and made a series of measurements of the RFI environment at the Goldstone-Venus development station. Described in this article are the receiving system used [ l ] , and the results of one low-sensitivity survey performed Febru- ary 16-24, 1987. Additional surveys in restricted frequency bands and with higher sensitivify have been carried out in the interim. These surveys have not been fully analyzed at this time, This RFI survey affort is expected to continue for

several more years. Efforts will be concentrated on establish- ing trends and on attaining sensitivity commensurate with levels that will be achieved by SETI.

II. The Radio Spectrum Surveillance System The RSSS used for the observations reported here consisted

of a log-periodic feed, a 1-meter parabolic antenna, a set of seven GaAs FET amplifiers (followed by transistor amplifiers) covering the spectrum from 1.0-10.4 GHz, and a Tektronix 494P swept spectrum analyzer. Figure 1 shows a block dia- gram of the receiving system. A Tektronix 4052A digital computer/controller automatically controls the antenna azimuth, noise diode, amplifier section, spectrum analyzer, and dual-floppy disk recorder. The spectral data are processed in the 4052A in near real time to determine whether a speci- fied power threshold has been exceeded anywhere in the

179

spectrum. The data describing such events are written on the floppy disks at the field station for further off-line analysis at JPL.

between 3:OO a.m. and 4:OO a.m. local time. Table 2 gives the parameters used in the survey.

The antenna and amplifiers are mounted atop the roof of Iv. Results a mobile trailer van. The spectrum analyzer, computer, and recording equipment are mounted in a rack inside the van. For the observations reported here, the van was parked at the Goldstone-Venus development station (DSS 13), about 30 meters south-west of the main control building. In this loca- tion, the van is approximately 150 meters east of the 26-meter antenna. Figure 2 shows a photograph of the site to illustrate the relative positions of the control building, the 26-meter antenna, and the RSSS.

A total of 37,589 events were detected in the survey. Fig- ure 4 shows a histogram of the number of events exceeding the threshold as a function of received power level. Note that the power levels for some interfering signals exceeded -90 dBm. These signals were traceable to local microwave transmitters. The relatively low number of signals detected with power levels less than -1 18 dBm is believed to be caused by the non- uniform sensitivity of the survey which varied by approxi- mately 7 dBm from one end of the band to the other due to increasing receiver noise temperature at higher frequencies. It has not been concluded that weaker signals are less prevalent than stronger signals; further work is planned to clarify the situation at lower power levels. Figure 5 shows the cumulative distribution of events as a function of minimum received power level.

Data recorded on the floppy disks are taken to JPL and copied onto a hard disk of a VAX 11/750 computer for further analysis. A commercially available database program, INGRES, is used to retrieve the data; custom software is used in conjunction with INGRES to perform parameter searches and generate various graphical displays.

111. Observations The observations were carried out in a fully automated

mode by pre-scheduling the controller t o scan the spectrum analyzer from 1 .O GHz to 10.4 GHz repeatedly at a resolution bandwidth of 10 kHz. The time required to complete a single- frequency scan was 40 minutes, and in total, 232 scans were made over the course of the survey. In order to expedite the survey, the antenna main beam remained motionless and pointed in the direction of the zenith instead of scanning the horizon. In the zenith orientation, the antenna gain resembles that of an omnidirectional antenna for radiation that arrives along the horizon. Assuming that the antenna gain in the direction of the horizon is the same as that from an omni- directional antenna, the effective area for the survey is approxi- mately 0.1 X h 2 , where X is the wavelength of the observation. The effective area for on-axis signals is approximately 0.5 m2. Power levels reported in this article assume an omnidirectional antenna.

Each frequency was observed a total of 232 times, with nearly uniform coverage with time of day. Figure 3 shows the distribution of observations (“looks”) with day of the week. Each day of the week is further divided into six 4-hour inter- vals starting at midnight. The weekdays Monday and Tuesday were observed most frequently because the survey began on a Monday and extended for nearly nine contiguous days. Sun- day was observed least frequently because of an interruption due to a full storage disk. Only the first 4-hour interval (mid- night to 4:OO a.m.) was observed on Sunday. A system noise temperature calibration was automatically performed daily

Figure 6 shows the probability of a signal exceeding the threshold as a function of frequency over the full frequency range of the survey, 1.0-10.4 GHz. The frequency resolution of the graph is approximately 10 MHz. This figure reveals that the probability is very nonuniform across the survey frequency range. The highest incidence of interference is in two frequency ranges: 1.0-3.0 GHz and 7.7-8.3 GHz. Figure 7 shows an enlargement of Fig. 6 in the frequency range 1.0-2.0 GHz. The frequency resolution of Fig. 7 is approximately 1 MHz. It can be seen from this figure that the “Water Hole” frequency range, 1.4-1.7 GHz, contains a significant amount of interfer- ence. but that the radio astronomy bands near 1.4 GHz and 1.6 GHz are relatively quiet. at least at the sensitivity achieved by this particular survey.

Figure 8 displays the probability of RFI events over the full 1 .O-10.4 GHz range as a function of day of week. As in Fig. 3, each day is divided into six 4-hour intervals. Note that no obser- vations were made between 0400 UT Sunday and 0400 UT Monday, which explains the lack of events during this time interval. This figure illustrates the pervasive nature of the inter- ference. RFI events were observed every day of the week, with little change in probability of occurrence with time of day.

Table 3 provides a list of the frequencies and approximate frequency ranges that have exhibited an interfering signal at least 10 percent of the time they were observed. A worst-case estimate of the fraction of the observed band which is obscured by RFI can be made by adding up all the frequency ranges in Table 3. From this crude summation, it is found that 7.21 MHz, or approximately 7.7 percent of the 9.4-GHz survey band- width is affected by RFI at least 10 percent of the time.

180

A more detailed analysis of the raw data, examining each channel individually rather than in the frequency ranges indi- cated in Table 3, is shown in Fig. 9. This figure depicts the percent of the observed band obscured by RFI at least 10 per- cent of the time as a function of a limiting received power. The solid curve spanning the range from -100 dBm to -1 16 dBm is derived from the survey. Since the sensitivity level at which SETI will be operating is close to -170 dBm, it is of great interest to determine experimentally the shape of this curve at lower power levels. The dashed curves shown in Fig. 9 are crude extrapolations to lower power levels.

Several statistical models which might be helpful in provid- ing a basis for understanding the present and future RFI sur- vey results were considered. In the first model it was assumed that the effective isotropic radiated power (EIRP) of each transmitter is identical. that the transmitters are randomly distributed in area. and that the transmitter frequencies are randomly distributed. These assumptions are certainly incor- rect since (1) transmitter frequency allocations are not ran- dom, ( 2 ) transmitter positions are generally clustered, and (3) EIRPs are not constant. Nevertheless. the model describes the extreme situation of extrapolating the local environment at Goldstone to more distant regions, assuming the areal density of transmitters is constant. For this model, the frac- tion of the survey band obscured, f, as a function of the received power, P, is given by the expression:

where k is a constant. The constant, k , is a function of the areal density of transmitters. the fractional bandwidth covered by each transmitter, the EIRP of each transmitter. and the collection area of the receiving system. The dashed curve in Fig. 9 shows a graph of this equation with k taken to be -136.7 dBm (note: convert power from dBm to Watts when using the equation). This equation thus predicts that 63 per- cent of the band is obscured at the -136.7-dBm level. Prelimi- nary results of a subsequent survey. not reported here. show that the interference is much less than this model predicts. The spatial distribution could just as well have been ignored in this model. and the spectral power density assumed to arrive at the same result. In reality. some combination of areal density and spectral power distribution are needed for a complete description.

A simple estimate of a lower bound for this model can be made by assuming that the areal density of transmitters goes

abruptly to zero at some arbitrary radius. This condition might exist if the detectable transmitters were clustered near Gold- stone, or if distant transmitters were obscured by the horizon. This lower bound model can be expressed by the pair of equations:

k -_ f = I - e P>P,

The locus of points defined by these equations is shown by the two dotted lines in Fig. 9. Each assumes the same value of k as above but different minimum power cutoffs. One is drawn for a value of P,,, = -125 dBm, and the other for a value of P,,, = -114 dBm.

It is impossible to draw any firm conclusions about these models until more sensitive survey data are available. The models are highly simplified and do not consider the reality of more than one population of transmitters. For example, the sensitivity of this survey was too low to detect signals from geosynchronous satellites. We emphasize that any attempt to extrapolate these data to the power levels of interest to SETI is considered to be highly uncertain because of the more than 50-dBm power level difference between the experimental data and the SETI power regime.

V. Conclusions It is concluded from this low-sensitivity, broad-band survey

that most of the strong (> -116 dBm) RFI in the 1-10 GHz band occurs in relatively few bands. Nearly 1 percent of the entire band shows interference a t the -116-dBm level or stronger with a probability of occurrence 2 1 0 percent. The interfering signals do not appear t o show a strong dependence on time of day or on day of the week. Interfering signals from satellites will probably show up at power levels significantly less than were achieved in this survey. Surveys at approxi- mately 30 dB better sensitivity can be achieved with the cur- rent RSSS in a reasonable time through (1) use of a discone antenna to provide higher sensitivity along the horizon, and (2) observing with a more narrow bandwidth. Observations are currently being made with higher sensitivity using both of these techniques. Surveys at even greater sensitivity, approach- ing those achieved by SETI in the sidelobes, must await more sensitive systems.

181

Ref e rence

[ 11 B. Crow, A. Lokshin, M . Marina, and L. Ching, “SET1 Radio Surveillance System,” TDA Progress Report 42-82, vol. April-June 1985, Jet Propulsion Laboratory, Pasadena, California, pp. 173-184, August 15, 1985.

I 182

Table 1. All-sky survey parameters

Spatial coverage Frequency range

Duration Frequency resolution Instantaneous bandpass

Sensitivity Polarization Signals

Entire celestial sphere 1-1 0 GHz inclusive and higher frequency - 6 years - 30 Hz = 250 MHz

< 10-23 G W m - 2 Simultaneous dual circular Primarily CW with natural radio

spot bands

astronomy I ~ I I O U [

Table 2. Survey 1 parameters

Start date Stop date Number of sweeps Antenna orientation Frequency range Resolution Time constant Baseline level Threshold

February 16,1987 February 24, 1987 232 zenith

10 kHz 5 x sec -133 dBm at 1 GHz, -126 dBm at 10 GHz 10 dB above the baseline level

1 .O-10.4 GHz

Table 3. Frequencies with probability of interference B 10 percent

Center Frequency Probability of RFI, MHz of RFI R F Range Affected Band

Allocationa

1000.015 0.6 < 10 kHz wide AN 105 1.685 0.4 < 10 kHz wide AN 1085.420 0.5 1085.4 MHZ-1085.44 MHz AN 11 34.600 0.1 < 10 kHz wide AN 1165.975 0.6 1165.94 MHz-1166.11 MHz AN 1302.515 0.4 1302.5 MHz-1302.53 MHz A 1310.850 1 .o < 10 kHz wide AN 1723.41 0 0.2 < 10 kHz wide F&M 1723.525 0.6 < 10 kHz wide F&M 1723.640 0.2 < 10 kHz wide F&M 1724.5 25 1 .o 1724.5 MHz-1724.55 MHz F&M 1905.020 0.6 < 10 kHz wide F&M 2000.030 0.1 < 10 kHz wide F&M 21 10.850 0.1 < 10 kHz wide AF&M 21 12.420 0.2 < 10 kHz wide F&M 21 15.225 1 .o < 10 kHz wide F&M 2605.020 0.5 < 10 kHz wide BS 7786.785 0.9 7784.39 MHz-7789.18 MHz F 7832.015 0.2 < 10 kHz wide F 7879.970 0.5 7879.95 MHz-7879.99 MHz F 7884.960 0.7 7884.93 MHz-7884.99 MHz F 7889.970 0.3 7889.96 MHz-7889.98 MHz F 7934.430 0.3 7934.15 MHz-7934.71 MHz F&MS 7934.960 0.3 7934.89 MHz-7935.03 MHz F&MS 7935.110 0.1 < 10 kHz wide F&MS 7935.355 0.4 7935.25 MHz-7935.46 MHz F&MS 7935.770 0.2 7935.69 MHz-7935.85 MHz F&MS 8080.050 0.8 8079.73 MHz-8080.37 MHz F&MS 8179.91 0 1 .o 8179.88 MHz-8179.94 MHz F&MS 8260.095 0.4 8260.06 MHz-8260.13 MHz F&MS 8359.910 0.5 8359.89 MHz-8359.93 MHz F&MS

aA = Aeronautical AN = Aeronautical Navigation BS = Broadcasting Satellite F&MS = Fixed & Mobile Satellite

(Partial listing from “Tables of Frequency Allocations and Other Extracts” from Manual of Regulations and Procedures for Federal Radio Frequency Management, Superintendent of Documents, Washington, D.C., January 1984.)

F = Fixed F&M = Fixed& Mobile

183

ANTENNA 1-m DlAM \ SOURCE

NOISE DIODE ON

OFF O.O1--18GHz

184

REMOTE ON/OFF

1-10dB + ROTARY JOINT -

I

-

1 5 0 il

-

7 GaAs FET AMPS ANTENNA ROTATOR ASSEMBLY 1-1OGHz

60 FT COAX 60 FT COAX

1 I

Fig. 2. Photograph of RSSS mounted on mobile van at Goldstone-Venus development station.

0 60 FT COAX

I I

ROTATOR CONTROLLER

SPECTRUM R E LAY ANALYZER DRIVER TEKTRONIX 494P

FREQUENCY - COUNTER HP 59306A

DC POWER SUPPLY

1

DISK CONTROLLER LSI 11/2

CONTROLLER - TEKTRONIX 4052

EXTERNAL

HP 59309A

SET1 RADIO SPECTRUM SURVEILLANCE SYSTEM RACK

I I I I I I

NODATA

10

v) Y 0 s # E

U

a: 0

I 3 z

0

DAY OF THE WEEK (PACIFIC STANDARD TIME)

Fig. 3. Number of looks versus day of week for Survey 1, 1.0-10.4 GHz.

I I I I I I I I I

1 O O L -130 -125 -120

-T 1-T T

L

-115 -110

T

-105

232 LOOKS 37,588 EVENTS

T

-100 -95

RECEIVED POWER, dBm

-90 I Fig. 4. Histogram of number of RFI events as a function of received power for Survey 1,

1.0-10.4 GHz.

185

1 .o

0.8

> k m

0 n

n

6

0.6 - 2 LL

Lu 1

0.4 u)

0.2

0 1 L 3

~

-130 -120 -110 -100 -90 -80

P,. dBm

Fig. 5. Cumulative histogram of number of RFI events as a function of P, for Survey 1, 1.0-10.4 GHz.

I I

I 4 9 10

FREQUENCY, GHz

Fig. 6. Probability of RFI events as a function of frequency for Survey 1, 1.0-10.4 GHz.

186

0 1 .o

0 LT P

w 1 P 5 0.4 v)

0.2

$ll 1.2 1.4 1.5 1.6 1.7

FREQUENCY, GHz

1.8 1

Fig. 7. Probability of RFI events as a function of frequency for Survey 1, 1.0-2.0 GHz.

I I I 1 1 I I

NO DATA

MONDAY I TUESDAY I WEDNESDAY I THURSDAY I FRIDAY I SATURDAY I SUNDAY

DAY OF THE WEEK (PACIFIC STANDARD TIME)

Fig. 8. Probability of RFI events as a function of &hour slice of day of week for Survey 1, 1.0-10.4 GHz.

187

102

101

O W n 3 u

0 0 z

2 100

2 + 10-1

LL 0

z w V n w L

10-2

10-3

I I U I I I I \ 1

1 CONSTANT AREAL DENSITY MODEL \

P, = -125 dBrn \

\. \.

--_---------- P, = -1 14 dBrn -__------ CONSTANT AREAL DENSITY MODELS WITH CUTOFF A T SPECIFIED POWER LEVELS

I I I I I I I 1 I70 -160 -150 -140 -130 -120 -110 -100 -90

RECEIVED POWER, dBm

Fig. 9. Observed and extrapolated percent of bandpass obscured by RFI events with probability greater than 10 percent, 1.0-10.4 GHt.

188

I TDA Progress Report 42-96 October - December 1988

R. E. Hill (Ground Antenna and Facilities Engineering Section) has submitted the fol- lowing errata to his article “A New State Space Model for the NASA/JPL 70-Meter Antenna Servo Controls” that appeared in the Telecommunications and Data Acquisi- tion Progress Report 42-91, July-September 1987, November 15, 1987:

In Table 2 (page 253), the denominator, Ji , appearing in the equation f o r i , should be replaced with J B . The variable, x i , appearing in the equation for i2i+4 should be replaced with xl. The inequality symbol (#) appearing in the equation for i2i+4 should be replaced by a plus symbol (t).

In Table 3 (page 254), dashed lines are added to the linear system matrix to show the respective rows and columns corresponding to the flexible structure and alidade modes. The complexity of the changes to several elements of matrix F necessitate the reproduc- tion of the revised version in its entirety, below.

Also in Table 3 (page 255), the output vector HEe should appear:

H E , = [ l 0 0 0 . . . -1 0 0 0 01 Elevation only

In the Notes section of Table 3 (page 255), the following errata were submitted:

Is Should Be

a1 . . .a5

a0 t a l t . . . a N = 1

a1 . . . a 5

a,-, t u , t . . . a N = 1 a0 a0

4l 4 n Finally, in Table 4 (page 256), the term y should be - V ‘

189

F =

0 1 I

0 KN

0 - K 3

0 - -KG - KT KG I K l K 2

0 ' - 0 - 0 - JB JB I JB JB JB JB

I 0 1 1

I I FLEXIBLE MODES c

2 2 I* 2N COLUMNS m -W

0

2 1 W

0

2 2 W

0

W2 3

0

44

0

D l

J ,

-

0

O 2

J 2

0

D3

J3

-

0

DN

JN

-

C I -J

0 1

2 -D,

J 1 -w -

0 1

2 -D2

J2 -w -

2N h3WS 0 1

2 4 3

J 3 --3 -

0

2 -"N

1

-DN

JN

I I O I I I o I I I -KG l - I I

ALIDADE MODES -1 I I I

- 4 COLUMNS

I I I I I

I I I I I I I I I I I I I I I I I

I

I

I

0 l I I i

' K A l - K A 2 I

JA2 O 1 I

JM

0

Note: The matrix F is defined in a generalized form with a variable N , corresponding to the number of flexible modes (not including alidade modes) and is illustrated for the special case of N = 4. With the deletion of the rows and columns corresponding to the alidade, the generalized matrix structure is applicable to the azimuth axis. It should be noted that the F matrix structure is similar to a variant of the Jordan canonical form in that the damping coefficients appear along the diagonal and the squared natural frequencies appear along the subdiagonal with unit elements along the superdiagonal.

190